Multidimensional Cell-Free DNA Fragmentomic Assay for Detection of Early-Stage Lung Cancer

Area of Science:

• Oncology
• Genomics
• Bioinformatics

Background:

• Cell-free DNA (cfDNA) analysis shows promise for early lung cancer detection, improving patient survival.
• Existing cfDNA methods lack sufficient sensitivity for early-stage tumors, limiting clinical application.

Purpose of the Study:

• To develop an accurate and cost-effective method for early-stage lung cancer detection.
• To integrate cfDNA fragmentomics with machine learning for enhanced diagnostic capabilities.

Main Methods:

• Whole-genome sequencing of plasma cfDNA from 782 participants (432 cancer, 350 healthy).
• Development and comparison of various machine learning models and cfDNA features.
• Validation of the optimal model in three independent cohorts.

Main Results:

• A stacked ensemble model integrating five cfDNA features and algorithms achieved high performance.
• Sensitivities ranged from 84.7% to 92.5% with specificities above 94.2% across validation cohorts.
• The model demonstrated high sensitivity for early-stage (Stage I: 83.2%) and small tumors (<1 cm: 85.0%).

Conclusions:

• A novel stacked ensemble model using cfDNA fragmentomics significantly enhances early-stage lung cancer detection.
• This approach offers superior sensitivity and specificity, paving the way for earlier diagnosis and improved patient outcomes.